WO2011050454A1 - Appareil et procédé de segmentation d'imagerie osseuse - Google Patents

Appareil et procédé de segmentation d'imagerie osseuse Download PDF

Info

Publication number
WO2011050454A1
WO2011050454A1 PCT/CA2010/001680 CA2010001680W WO2011050454A1 WO 2011050454 A1 WO2011050454 A1 WO 2011050454A1 CA 2010001680 W CA2010001680 W CA 2010001680W WO 2011050454 A1 WO2011050454 A1 WO 2011050454A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
bone
input
compounded
contour
Prior art date
Application number
PCT/CA2010/001680
Other languages
English (en)
Inventor
Fréderic LAVOIE
Said Benameur
Original Assignee
Eiffel Medtech Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Eiffel Medtech Inc. filed Critical Eiffel Medtech Inc.
Priority to CA2778599A priority Critical patent/CA2778599C/fr
Publication of WO2011050454A1 publication Critical patent/WO2011050454A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/168—Segmentation; Edge detection involving transform domain methods
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/50—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications
    • A61B6/505—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications for diagnosis of bone
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10081—Computed x-ray tomography [CT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10116—X-ray image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10132—Ultrasound image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20021—Dividing image into blocks, subimages or windows
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20048—Transform domain processing
    • G06T2207/20052—Discrete cosine transform [DCT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30008—Bone

Definitions

  • This specification relates to the field of medical imagery, and more particularly to image analysis, border detection and segmentation.
  • Prior art image segmentation methods have been applied in medical imaging to detect micro-calcifications in mammogram, to classify cancerous tissues in MRI images, to evaluate bone structures from X-ray images, to detect lesions in images of organs, or to segment and to classify tumors shown in ultrasound images.
  • prior art methods present several shortcomings, such as when segmenting bones represented by ultrasound images.
  • Prior art methods also present issues when attempting to create a bone structure model using various types and qualities of bone images. For example, errors may occur from time to time depending on image quality (e.g. detecting a bone contour where in fact the image shows a tone variation representative of different body tissues).
  • image quality e.g. detecting a bone contour where in fact the image shows a tone variation representative of different body tissues.
  • typical techniques are rather complex and consuming in terms of time and processing resources. Such drawbacks can become quite irritating for a surgeon during a surgical procedure for example.
  • the apparatus and method herein described are applicable in creating electronic bone structure models from images taken on a specific patient.
  • the models thus created are customized to a given patient and provide improved assistance to a surgeon during surgery for example.
  • the herein described apparatus and method are useful in computer-assisted medical applications.
  • images of a patient's member can be automatically processed during a surgical procedure as per the present description, to determine three-dimensional characteristics of a bone. Bone contours and other characteristics thus determined can be displayed to a surgeon in a given spatial coordinate system for example, in order to assist in a medical procedure.
  • the presently described apparatus and method thus alleviate the cumbersome tasks of analyzing multiple images of various types, forms and quality levels, and comparing them to one another before, during and/or after a medical procedure. Better monitoring and diagnosis of a patient's bone condition is also possible from the presently described apparatus and method.
  • pixel is intended to refer to a unit picture element which forms part of a digital image.
  • pixel is meant to describe the smallest electronic value used by an electronic apparatus in representing a unit point in an image.
  • region is intended to refer to a block of adjacent pixels of substantially similar tone/color values.
  • border is intended to refer to a set of contours, which once joined in a same coordinate system, define a closed space in two or three- dimensions. It should be noted that the prior art often refers to this definition as corresponding to a contour.
  • the present specification provides an image segmentation method for recovering a contour of a bone from an input image of the bone.
  • the method comprises receiving the input image at a processing device; in the processing device, applying in parallel at least three image processing functions to the input image, to obtain at least three resulting images indicative of respective features of the input image, at least one of the at least three image processing functions pertaining to a spatial domain, and at least another one of the at least three image processing functions pertaining to a frequency domain; in the processing device, combining the at least three resulting images together to form a single compounded image, the compounded image identifying at least two regions based on the respective features, one of the at least two regions corresponding to the bone; in the processing device, identifying the contour of the bone based on the at least two regions of the compounded image; and outputting an output image for display, the output image being based on the compounded image and comprising the contour identified.
  • an image segmentation apparatus for recovering a contour of a bone from an image of the bone.
  • the apparatus comprises an input device for receiving the image of the bone; an output device for outputting an output image; a processing device; and a memory device in operative communication with the processing device and the input device.
  • the memory device comprises instructions for implementing the processing device to: apply in parallel at least three image processing functions to the image, to obtain at least three resulting images indicative of respective features of the image, at least one of the at least three image processing functions pertaining to a spatial domain, and at least another one of the at least three image processing functions pertaining to a frequency domain; combine the at least three resulting images together to form a single compounded image, the compounded image identifying at least two regions based on the respective features, one of the at least two regions corresponding to the bone; identify the contour of the bone based on the at least two regions of the compounded image; and output the output image to the output device, the output image being based on the compounded image and comprising the contour identified.
  • an image segmentation apparatus for recovering a contour of a bone from an image of the bone.
  • the apparatus comprises at least three image processing units each receiving the image of the bone, the at least three image processing units processing the image to obtain at least three respective results indicative of respective features of the image, at least one of the at least three image processing units processing the image in a spatial domain, and at least another one of the at least three image processing units processing the image in a frequency domain; a combining unit in operative communication with each one of the at least three image processing units for receiving the at least three respective results, and for combining the at least three respective results together to form a compounded result, the compounded result being indicative at least two regions as defined by the respective features, one of the at least two regions corresponding to the bone; and a bone detecting unit in operative communication with the combining unit, for identifying the contour of the bone based on the at least two regions of the compounded result.
  • FIG. 1 is a schematic illustration of an apparatus for recovering a contour of a bone from an image of the bone, in accordance with an embodiment
  • FIG. 2 is a detailed schematic illustration of another apparatus for recovering a contour of a bone from an image of the bone, in accordance with an embodiment
  • Fig. 3 is a flow chart illustrating a method for recovering a contour of a bone from an image of the bone, in accordance with an embodiment
  • Fig. 4 is an example of a gray-scale ultrasound input image of a bone input in accordance with an embodiment
  • Fig. 5 is an example of a processed image resulting from the application of a first function on the input image of Fig. 4, in accordance with an embodiment
  • Fig. 6 is an example of a processed image resulting from the application of a second function on the input image of Fig. 4, in accordance with an embodiment
  • Fig. 7 is an example of a processed image resulting from the application of a third function on the input image of Fig. 4, in accordance with an embodiment
  • Fig. 8 is an example of a processed image resulting from the application of a fourth function on the input image of Fig. 4, in accordance with an embodiment
  • FIG. 9 is an example of a compounded image resulting from the fusion of the processed images of Figures 5 to 8, in accordance with an embodiment
  • Fig. 10 is the compounded image of Fig. 9, after a first stage leading to the detection of a contour therefrom, in accordance with an embodiment
  • Fig. 11 is the compounded image of Fig. 10, after a next stage leading to the detection of a contour therefrom, in accordance with an embodiment
  • Fig. 12 is an example of an output image with a border, in accordance with an embodiment .
  • Fig. 13 is an example of an output image with a contour, in accordance with an embodiment.
  • FIG. 1 there is illustrated an apparatus 100 for recovering a contour of a bone from an input digital image of the bone.
  • the apparatus 100 has an input device 102, a display device 104, a memory unit 106 and a processing device 108.
  • the input device 102 receives an input signal representative of an input image of the bone.
  • the input signal has values of entries in an input image matrix representative of the image of the bone.
  • the input image can be any type of image: gray-scale, color, two- or three-dimensional.
  • Various types of medical images are also applicable, such as MRI, X-Ray images, computed tomography (CT), and ultrasound images.
  • the output device 04 outputs an output image, or an output image matrix representation of the output image, so as to provide the contour of the bone that is recovered by the apparatus 100 from the input image.
  • the output device 104 can be any type of device which uses the recovered bone contour as an input to another image-related process. Alternatively, the output device is simply a display device.
  • the memory 106 is in operative communication with the processing device 108 and receives the input image from the input device 102.
  • the memory 106 also stores instructions, which once run by the processing device 108, implement the processing device 108 to perform a series of tasks on the input image.
  • the input image(s), output image(s) and processed images obtained from various processing steps are also optionally stored in the memory device 106.
  • Coded instructions from the memory 106 instruct the processing device 108 to apply at least three image processing functions to the input image in a parallel fashion (e.g. each applied to the same input image).
  • the functions are applied independently and separately from each other on the original input image.
  • the functions are distinct from each other and permit the finding of different features of the image. Applying these three functions (or more; e.g. F functions, where F>3) to the input image results in F results, each representative of a single resulting image (i.e., F resulting images).
  • the F resulting images are each indicative of at least one feature difference between blocks of pixels, or individual pixels, of the input image.
  • the F functions are chosen such that at least one is applied in the spatial domain and at least another one is applied in the frequency domain.
  • Each one of the F functions permits the identification of a specific, distinct feature of the input image, be it related to texture, tone or visual structure. For example, a first function may be chosen to determine a tone distribution of the image; a second function may be chosen to determine a texture characteristic of the image; while a third function may be chosen determine a visual significance of a feature of the input image. If more than three functions are used, a fourth function may be chosen to determine a spatial or structural feature from the input image. Many other types of functions are optionally added and applied to the input image in a similar, parallel fashion, to distinguish more features of the input image.
  • the processing device 108 can be a single processing unit which performs the functions F one after each other in time, but each on the input image; or a single parallel- processing unit which is adapted to perform the functions F concurrently.
  • a single non-parallel processing unit is used, each result obtained from performing one of the functions F is stored in the memory device 106 prior to continuing with another one of the functions F.
  • the processing device 108 can also be a combination of various types of processing units.
  • F resulting images More coded instructions from the memory 106 instruct the processing device 108 to combine the F results (also referred as F resulting images) together to form a single compounded image.
  • F results also referred as F resulting images
  • Various image compounding techniques can be used, such as a K-means compounding technique.
  • the pixels are redistributed in the k classes according to their characteristics in each one of the resulting images.
  • the compounded image is indicative of all of the feature information found by applying each one of the functions.
  • the memory device 106 also implements the processing device 108 to identify the contour of the bone based on the compounded image. For example, a set of larger sized unified regions are located on the compounded image, and gradients associated with regions of the input image corresponding to the set of larger sized unified regions, are obtained in order to identify the contour.
  • the processing device 108 is also optionally instructed to form a line demarcation representative of the contour using a contour recovery process applied to the gradients.
  • the coded instructions from the memory 106 also instruct the processing device 108 to output the output image to the output device 104.
  • the output image is based on the compounded image and identifies the contour as per the indications provided in the compounded image.
  • the output image has the contour of the bone, or indicates its presence by having at least two regions each with a substantially unified tone, sufficiently different from the other to create a contour demarcation. More than one recovered contour can be in the output image.
  • the output image optionally has a contour formed by all of the recovered contours, and possibly a side of the image, connecting together to form a closed region of a substantially unified tone.
  • the memory device 106 can be used to store the input image as well as the compounded image and the output image before it is displayed on the output device 104.
  • Other instructions can be coded and stored in the memory 106 to implement the processing device 108 to scale the input image from an original size to of a reduced size prior to applying the image processing functions F.
  • Image scaling permit greater rapidity of execution and lower use of available processing power and resources.
  • the scaled-down image (or a matrix representation of a scaled-down image) is optionally stored by the memory device 106.
  • the processing device is optionally instructed to scale a number or all of the F resulting images; or alternatively the compounded image, back to the original size of the input image. This way, the output image has the same size as the original input image.
  • FIG. 2 there is illustrated an apparatus 200 according to an embodiment in which at least three different image processing units, 202, 204, 206, 208, .., (F), are used separately and in parallel to perform different image processing tasks on the input image entered at the input device 102.
  • the input image is an image matrix representation 210 carried over a communication signal for example, or residing in a local or remote memory unit (not shown) or other support medium such as a hard drive, or a removable disk.
  • the input image 210 is communicated by the input device 102 to each one of the processing units 202, 204, 206, 208, and (F), via an optional de-scaling unit 212.
  • the input device 102 also communicates the input image 210 to a contour detecting unit 214.
  • each one of the processing units, 202, 204, 206 and 208, (F) applies a single one of the image processing functions F mentioned above in reference to Fig. 1 .
  • each of the units 202, 204, 206, 208, (F) outputs a respective result as also mentioned above in relation to Fig. 1.
  • a combining unit 2 6 combines together the F results each outputted by respective processing units, 202, 204, 206, 208,... (F) to obtain and output a compounded result (or corresponding compounded image) representative of all of the feature differences of each one of the F results obtained from each of the processing units (F).
  • the compounded image has multiple regions, one of which is associated with the bone.
  • the contour detector 214 receives the compounded result from the combining unit 216 and identifies the contour of the bone therefrom, as per a process described hereinabove in relation to the processing unit 108 of Fig. 1.
  • the output device 104 receives the input image 210 from the input device 102, and the identified contour from the contour detector 214.
  • the output device 104 outputs an output image based on the compounded image and the contour of the bone.
  • the output image is digital and represented by an output image matrix 220.
  • the optional de-scaling unit 212 may scale down the input image to send a scaled-down input image to only one or more of the processing units (any one or more of units 202 to F).
  • an optional re-scaling unit 218 re-scales the resulting image(s) outputted by the respective only one or more of the processing units such that all of the re-scaled resulting images sent to the combining unit 216 are of the same size prior to being combined together.
  • the re-scaling unit 218 is after the combining unit 216. This case is feasible when all the processing units 202 to F received a scaled-down input image (i.e. the sizes of all of the resulting images are equivalent).
  • the re-scaling unit 218 is in operative communication only with the combining unit 216 and the output device 104, while the contour detector 214 communicates the contour directly to the output device 104.
  • the re-scaling unit 2 8 is embodied as multiple re-scaling units (or F re-scaling units) each operatively coupled to, or forming part of, each one of the processing units 202, 204, 206, 208, ...(F). Similarly for the de-scaling unit 212.
  • the De-scaling and Re-scaling units 212 and 218 are optional and used to reduce the input image in size prior to processing, as well as to increase the output image's size to correspond to the original size of the input image.
  • a scaling factor is communicated between the scaling units 212 and 218.
  • unit 212 alternatively communicates the scaled-down input image to the contour detector 214.
  • each one of the processing units 202, 204, 206, 208, (F) are electronic filtering devices which may be implemented as programmable logic devices, for example, programmed to have a specific response equivalent to the function they are each specifically meant to apply to the input image.
  • both of the apparatuses 100 and 200 of Figures 1 and 2 optionally comprise an image acquisition device or a scanning device (not shown) operatively coupled to the input device 102, for either taking an image of a bone or scanning an image of a bone that is not already available in digital format.
  • a flow chart illustrates a method 300 for recovering a contour of a bone from an input image of the bone.
  • step 302 the input image is received at an input of a processing device.
  • the input image is first acquired by a digital image acquisition device.
  • Step 304 is optional and involves scaling the input image from an original size of the input image to of a reduced size (e.g. de-scaling), to obtain a scaled- down (or de-scaled) image.
  • a reduced size e.g. de-scaling
  • a processing device separately applies, in parallel, at least three image processing functions to the input image (or to the scaled-down image), to respectively obtain at least three resulting images.
  • Each one of the resulting images is indicative of respective features (or distinct differences in texture, tone or any other aspect) between blocks of pixels, or individual pixels, of the input image.
  • At least one of the image processing functions pertains to a spatial domain, while at least another one of the image processing functions pertains to a frequency domain.
  • step 308 is optional and involves scaling the resulting images back to the original size of the input image (e.g. re-scaling).
  • the re- scaling uses a scaling factor which was used in step 304 to de-scale the input image.
  • Various types of multi-scaling techniques are usable to achieve steps 304 and 308. If step 304 is achieved, step 308 is generally performed as well.
  • step 310 the at least three resulting images are combined together in the processing device to form one, single compounded image.
  • the compounded image identifies at least two regions in light of the respective features; one of the regions corresponds to the bone.
  • step 312 the contour of the bone is identified in the processing device based on the compounded image, or more specifically the regions identified in the compounded image.
  • an output image is outputted from the processing device for display to an output device.
  • the output image is based on the compounded image and comprises the contour identified in step 312.
  • the contour may be displayed as a line demarcation on the compounded image.
  • the output image represents the bone of the input image segmented therefrom.
  • step 308 can also be performed after step 310 if all of the resulting images from step 306 are equivalent in size. This is the case, for example, if a scaled-down version of the input image is provided by step 304 and the image processing functions applied thereto.
  • step 310 optionally involves redistributing the pixels in the k classes according to their characteristics in each one of the resulting images, using a K-means algorithm.
  • step 312 a sorting of the various regions detectable from the compounded image is performed according to their respective sizes. Once this is achieved, a number of the regions having the largest sizes are identified and located on the input image. Tone gradients associated with each one of the number of regions having the largest sizes are determined from the input image (or its scaled-down version). The contour is then detected based on these tone gradients. A highest tone gradient is typically associated with the presence of a bone contour since bones are typically lighter in tone than their surroundings. Such tone gradient assumptions can be reversed or adapted for specific input images. Any suitable contour generating process can be used in step 310 to generate an image with a contour demarcation based on the specific tone gradients associated to regions defining the contour.
  • the method 300 is adaptable to recover a contour from any type of digital image, in two or three-dimensions.
  • the above-described method 300 is optionally applied to multiple input images which may each represent a bone from varying angles and depths. From these images, a bone contour can be recovered in a three- dimensional coordinate system (x, y and z, where z is a depth axis).
  • the multiple two-dimensional input images are processed to find a contour in each one of the images. Once all the contours are recovered for varying depths, they are translated into a same coordinate system, or compared with respect to each other in terms of their (x, y) coordinates and their depth (z). A three-dimensional output image is thus outputted showing a three-dimensional border of the bone.
  • This method is applicable to various medical domains, in which various types of computer models of a patient's anatomy are constructed based on data gathered directly from the patient. This way, personalized computer models of patients' bone structures can be created for use in medical treatment, surgery procedures and other medical interventions.
  • three-dimensional contour recovery can be achieved during surgical intervention.
  • An imaging probe is used to acquire images of the bone, while a depth indicator such as a needle is used to gather depth (z) coordinate values associated with each image being acquired.
  • step 306 optionally involves, for example, passing an image matrix A representative of the input image under analysis, to various distinct image processing devices.
  • the functions may be chosen in terms of desired feature detection.
  • the functions pertaining to the spatial domain may include functions for calculating a tone distribution histogram of pixels in the input image, texture characterization functions which may or may not be based on statistical methods, and other spatial filters which operate on pixel values as well as their positions.
  • the application of each distinct function generates one or more resulting images which are each indicative of a specific aspect and/or feature of the input image. For example, if tone distribution information of the input image is obtained by applying a quantized histogram function, the resulting image or set of resulting images are formed from the obtained tone distribution information.
  • K is the Kronecker delta function and is a normalization constant ensuring that:
  • This function concentrates on the spatial statistics involved in associating texture features with tone levels of an image.
  • the below provided example is for a gray scale, two-dimensional image.
  • the method and equation presented herein is however adaptable to color images and three-dimensional images, either in gray scale or color.
  • One method of identifying texture features in an image is by applying a matrix function to obtain what is best known as the co-occurrence matrix.
  • Haralick proposed a set of second-order statistics in order to describe the co-occurrence function p of a gray scale image, also termed the gray-level cooccurrence matrix G.
  • Such a matrix is square with dimension N g , where N g is the number of gray levels in the image.
  • the matrix G has a number of rows and columns each equivalent to a number of gray levels in a particular image being analysed.
  • an matrix entry [i, j] of the gray-level co-occurrence matrix G is generated by counting a number of times a pixel with value i is adjacent to a pixel with value j; and then dividing the entire image matrix A representative of the image under analysis, by the total number of such comparisons made.
  • Each entry of the co-occurrence matrix G is therefore considered to represent the probability that a pixel with value / is found adjacent to a pixel of value j in the image analyzed.
  • the discrete cosine transform is applied to represent the image under analysis as a sum of sinusoids of varying magnitudes and frequencies. Once the DCT of an image is obtained, visually significant information about the image is distinguishable from its concentration in a few coefficients of the DCT.
  • a mathematical representation of the DCT is as follows:
  • the Gabor filter is a function in the spatial domain which is best described as a Gaussian function modulated by a sinusoidal curve; and may be mathematically represented as:
  • Fig. 4 is an example of a gray-scale ultrasound input image of a bone and surrounding tissue that could be received at step 302 of Fig. 3.
  • Fig. 5 through 8 each show different resulting images obtained after respectively applying four functions as per step 306 of Fig. 3.
  • Fig. 5 is an example of first processed image resulting from the application of a function comprising a quantified histogram function
  • Fig. 6 is an example of second processed image resulting from the application of a function comprising a cooccurrence matrix
  • Fig. 7 is an example of third processed image resulting from the application of a function comprising a discrete cosine transform (DCT);
  • Fig. 8 is an example of a fourth processed image resulting from the application of a function comprising a Gabor filter.
  • DCT discrete cosine transform
  • Fig. 9 shows a compounded image formed by fusing the resulting images of Figures 5 through 8, as per step 310 of Fig. 3.
  • Figure 9 has various regions.
  • Fig. 10 and Fig. 1 1 each show the compounded image of Fig. 9 at various stages during the contour identification process as per step 312 of Fig. 3. More particularly, Fig. 10 is the compounded image of Fig. 9 after detection of regions likely to contain the body to be detected; while Fig. 1 1 is the compounded image of Fig. 10 after detection of the regions containing the body to be detected.
  • Fig. 12 and Fig. 13 are examples of output images obtained based on the compounded image of Fig. 11 , as per step 314 of Fig. 3.
  • Fig. 12 shows an external border of the region containing the detected body; while Fig. 13 is an example of a contour, here an upper outer edge of the region containing the detected body.
  • Fig. 13 may be obtained from Fig. 12, or may be obtained prior to obtaining the entire external border of Fig. 12 (i.e. at a stage towards completing the border from a plurality of outer edges each forming contours).

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)

Abstract

L'invention porte sur un appareil et sur un procédé pour la récupération d'un contour d'un os à partir d'une image d'entrée de l'os avec ses tissus environnants. Le procédé comprend la réception de l'image d'entrée; l'application en parallèle d'au moins trois fonctions de traitement d'image sur l'image d'entrée pour obtenir au moins trois images résultantes indicatives de caractéristiques respectives de l'image d'entrée, au moins l'une des au moins trois fonctions de traitement d'image appartenant à un domaine spatial, et au moins une autre des au moins trois fonctions de traitement d'image appartenant à un domaine de fréquence; la combinaison des au moins trois images résultantes ensemble pour former une image composée, l'image composée identifiant au moins deux régions en fonction des caractéristiques respectives; l'identification du contour de l'os en fonction des au moins deux régions de l'image composée; et la délivrance en sortie d'une image de sortie pour l'affichage, l'image de sortie comprenant le contour identifié.
PCT/CA2010/001680 2009-10-30 2010-10-21 Appareil et procédé de segmentation d'imagerie osseuse WO2011050454A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CA2778599A CA2778599C (fr) 2009-10-30 2010-10-21 Appareil et procede de segmentation d'imagerie osseuse

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US12/609,839 US8644608B2 (en) 2009-10-30 2009-10-30 Bone imagery segmentation method and apparatus
US12/609,839 2009-10-30

Publications (1)

Publication Number Publication Date
WO2011050454A1 true WO2011050454A1 (fr) 2011-05-05

Family

ID=43921196

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CA2010/001680 WO2011050454A1 (fr) 2009-10-30 2010-10-21 Appareil et procédé de segmentation d'imagerie osseuse

Country Status (3)

Country Link
US (1) US8644608B2 (fr)
CA (1) CA2778599C (fr)
WO (1) WO2011050454A1 (fr)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8805035B2 (en) 2010-05-03 2014-08-12 Mim Software, Inc. Systems and methods for contouring a set of medical images
US8693744B2 (en) * 2010-05-03 2014-04-08 Mim Software, Inc. Systems and methods for generating a contour for a medical image
US10096120B2 (en) 2013-12-06 2018-10-09 Koninklijke Philips N.V. Bone segmentation from image data
US9558568B2 (en) 2014-06-27 2017-01-31 Siemens Healthcare Gmbh Visualization method for a human skeleton from a medical scan
EP3163536B1 (fr) * 2015-10-30 2021-12-08 Dassault Systèmes Compression d'un objet modélisé en trois dimensions
US10898079B2 (en) * 2016-03-04 2021-01-26 University Of Manitoba Intravascular plaque detection in OCT images
CN109949313A (zh) * 2019-05-17 2019-06-28 中科院—南京宽带无线移动通信研发中心 一种图像实时语义分割方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2004012584A2 (fr) * 2002-08-02 2004-02-12 Diagnostic Ultrasound Corporation Instrument a base d'ultrasons en 3d de mesure de structures remplies ou non de liquides
US6711282B1 (en) * 1999-10-29 2004-03-23 Compumed, Inc. Method for automatically segmenting a target bone from a digital image
US6983065B1 (en) * 2001-12-28 2006-01-03 Cognex Technology And Investment Corporation Method for extracting features from an image using oriented filters
EP2052674A1 (fr) * 2006-08-08 2009-04-29 Olympus Medical Systems Corp. Dispositif de traitement d'image medicale et procede de traitement d'image medicale

Family Cites Families (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5999639A (en) * 1997-09-04 1999-12-07 Qualia Computing, Inc. Method and system for automated detection of clustered microcalcifications from digital mammograms
US6137898A (en) * 1997-08-28 2000-10-24 Qualia Computing, Inc. Gabor filtering for improved microcalcification detection in digital mammograms
KR100338975B1 (ko) * 1999-09-14 2002-05-31 최은백, 이찬경 턱뼈의 치조신경 검출방법
US7123761B2 (en) * 2001-11-20 2006-10-17 Konica Corporation Feature extracting method, subject recognizing method and image processing apparatus
AU2003225508A1 (en) 2002-05-17 2003-12-02 Pfizer Products Inc. Apparatus and method for statistical image analysis
US7599579B2 (en) * 2002-07-11 2009-10-06 Ge Medical Systems Global Technology Company, Llc Interpolated image filtering method and apparatus
US7664298B2 (en) * 2003-03-25 2010-02-16 Imaging Therapeutics, Inc. Methods for the compensation of imaging technique in the processing of radiographic images
US7636461B2 (en) * 2004-02-05 2009-12-22 Koninklijke Philips Electronics N.V. Image-wide artifacts reduction caused by high attenuating objects in ct deploying voxel tissue class
US8064660B2 (en) * 2004-02-27 2011-11-22 National University Of Singapore Method and system for detection of bone fractures
CN101421745B (zh) * 2004-04-15 2016-05-11 美国医软科技公司 空间-时间肿瘤检测，分割和诊断信息提取系统及方法
US20060170679A1 (en) * 2005-02-01 2006-08-03 Hongwu Wang Representing a volume of interest as boolean combinations of multiple simple contour sets
JP2008532608A (ja) * 2005-03-11 2008-08-21 コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ ３次元超音波潅流画像のボリュームレンダリングシステム及び方法
US7660481B2 (en) * 2005-11-17 2010-02-09 Vital Images, Inc. Image enhancement using anisotropic noise filtering
US7664329B2 (en) * 2006-03-02 2010-02-16 Honeywell International Inc. Block-based Gaussian mixture model video motion detection
DE102006044189A1 (de) * 2006-09-20 2008-04-17 Siemens Ag Verfahren zum Bereitstellen von vielfältig verarbeiteten Bilddaten, Verfahren zum Verarbeiten vielfältiger Bilddaten und Röntgenbildsystem
EP1916624B1 (fr) * 2006-10-25 2016-11-23 Agfa HealthCare NV Procédé de segmentation d'une image médicale numérique
US8184915B2 (en) * 2006-12-04 2012-05-22 Lockheed Martin Corporation Device and method for fast computation of region based image features
US7881540B2 (en) * 2006-12-05 2011-02-01 Fujifilm Corporation Method and apparatus for detection using cluster-modified graph cuts
WO2008143849A2 (fr) * 2007-05-14 2008-11-27 Historx, Inc. Séparation en compartiments par caractérisation de pixel utilisant le regroupement de données d'image
DE102007053957A1 (de) 2007-11-09 2009-05-14 Henkel Ag & Co. Kgaa Stylingmittel mit hohem Haltegrad bei Feuchtugkeit IV
US8351666B2 (en) * 2007-11-15 2013-01-08 General Electric Company Portable imaging system having a seamless form factor
EP2362350B1 (fr) * 2008-05-08 2018-03-14 Oslo Universitetssykehus HF Normalisation automatisée de cartes de volume sanguin cérébral
US8467606B2 (en) * 2011-08-25 2013-06-18 Eastman Kodak Company Method for segmenting a composite image

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6711282B1 (en) * 1999-10-29 2004-03-23 Compumed, Inc. Method for automatically segmenting a target bone from a digital image
US6983065B1 (en) * 2001-12-28 2006-01-03 Cognex Technology And Investment Corporation Method for extracting features from an image using oriented filters
WO2004012584A2 (fr) * 2002-08-02 2004-02-12 Diagnostic Ultrasound Corporation Instrument a base d'ultrasons en 3d de mesure de structures remplies ou non de liquides
EP2052674A1 (fr) * 2006-08-08 2009-04-29 Olympus Medical Systems Corp. Dispositif de traitement d'image medicale et procede de traitement d'image medicale

Also Published As

Publication number Publication date
US8644608B2 (en) 2014-02-04
CA2778599C (fr) 2019-06-18
CA2778599A1 (fr) 2011-05-05
US20110103654A1 (en) 2011-05-05

Similar Documents

Publication Publication Date Title
CN110599508B (zh) 基于人工智能的脊柱影像处理方法及相关设备
Lim et al. Introducing Willmore flow into level set segmentation of spinal vertebrae
CA2778599C (fr) Appareil et procede de segmentation d'imagerie osseuse
EP2807630B1 (fr) Traitement et affichage d'une image des seins
US20220375079A1 (en) Automatically segmenting vertebral bones in 3d medical images
CN112116004B (zh) 病灶分类方法及装置、病灶分类模型的训练方法
Garcia-Hernandez et al. Analysis of the impact of digital watermarking on computer-aided diagnosis in medical imaging
Hille et al. Vertebral body segmentation in wide range clinical routine spine MRI data
CN110288698B (zh) 基于mri的半月板三维重建系统
TWI684994B (zh) 脊椎影像註冊方法
US20180365876A1 (en) Method, apparatus and system for spine labeling
Rathi et al. Analysis of watermarking techniques for medical images preserving ROI
KR102182357B1 (ko) Ct 영상 내 간암 영역 기반으로 3차원 분석을 하는 수술 보조 장치 및 방법
Lu et al. Texture analysis based on Gabor filters improves the estimate of bone fracture risk from DXA images
US10307124B2 (en) Image display device, method, and program for determining common regions in images
Myint et al. Effective kidney segmentation using gradient based approach in abdominal CT images
Yang et al. Fusion of CT and MR images using an improved wavelet based method
JP2008178666A (ja) 画像処理方法及び装置
EP3577630B1 (fr) Détection de régions à faible contenu d'informations dans des images numériques à rayons x
Sigit et al. Automatic detection brain segmentation to detect brain tumor using MRI
Koh et al. Automated boundary extraction of the spinal canal in mri based on dynamic programming
Pandey et al. A Framework for Mathematical Methods in Medical Image Processing
An et al. Enhancement of CT image using image fusion
US10467753B1 (en) Methods and apparatus for slicing three dimensional medical images along curved anatomical surfaces
Verma et al. Segmentation and Analysis of Diaphragm from Ultrasound Images

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 10825885

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2778599

Country of ref document: CA

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 10825885

Country of ref document: EP

Kind code of ref document: A1